Steps

1. Confirm that implementation scope, approach, and trade-offs are agreed in the conversation. A plan approved through the issue skill counts as agreed. If agreement is missing, run the grill-me skill and reach it before continuing.

2. Branch check: ensure you are in a Claude Code–managed worktree on a topic branch (Branch & Worktree Policy in the root CLAUDE.md). Fresh worktree setup: yarn install, then NX_WORKSPACE_ROOT_PATH=<worktree-absolute-path> yarn build.

3. Read the governing constraints before writing code: the package-level CLAUDE.md of every package you will touch, plus any matching .claude/rules/*.md.

4. Implement exactly what was agreed, test-first: add failing spec tests that define the expected behavior (never throwaway reproduction scripts), then make them pass.

5. /code-review medium (user-executed): /code-review cannot be invoked by Claude. Present the following to the user, ask them to run it, and wait here for the results. Then fix all findings.

   /code-review medium
   

   Writing a substitute review yourself, or skipping this step because you cannot invoke it, is forbidden.

6. Run the qa-engineer skill; fix all findings.

7. Run the product-manager skill; fix all findings (includes documentation consistency — JSDoc placement, comment policy).

8. Run yarn lint; fix all errors.

9. Run yarn test; fix all failures.

10. Commit following the git skill.

11. Create the PR following the pr skill.

Rules

• Fix-all scope: code-level fixes, added tests, and documentation fixes are applied on the spot. Findings that imply a spec change or scope expansion require the user's approval first.
• Loop back: if any step causes a major change of direction or a large code change, restart from step 5.
• No skipping steps: step 5 in particular — being un-invocable by Claude is not a reason to skip it; asking the user and waiting is the correct path. Skip a step only when the user explicitly says to.
• Strict scope: never expand beyond the agreed scope on your own.